TileGenerator: Skipping empty cluster

Hello,

I am working in PIX4Dmapper with images captured from a camera mounted at the rear of a vehicle and pointing downward toward the road.

My dataset does not contain EXIF metadata. For each image, I only have:

  • X, Y, Z coordinates
  • Roll
  • Pitch
  • Heading

I manually defined the camera model using the internal calibration parameters of my camera/lens system.

Camera information:

  • Camera: Matrix Vision mvBlueCOUGAR-X-1012bC
  • Lens: Computar LM8JC10MV0828-MPY2
  • Focal length: 8.5 mm
  • Image size: 4096 x 3008
  • Sensor size: 14.1312 mm x 10.3776 mm

Internal calibration used:

  • Ck: -8.31463
  • XH: 0.02678
  • YH: -0.00295
  • A1: -1.45518E-003
  • A2: 1.77331E-005
  • A3: 0
  • B1: 0
  • B2: 0
  • C1: 0
  • C2: 0

In the Quality Report:

  • 22 out of 24 images are calibrated
  • 2 images are disabled
  • Matching is yellow with a median of 911.422 matches per calibrated image
  • Overlap appears low in some areas
  • Absolute camera position and orientation uncertainties: “Uncertainty computation failed”

The processing fails during dense processing with these errors:

  • [Warning]: TileGenerator: Skipping empty cluster
  • [Error]: Dense matching failed
  • [Error]: Dense Processing failed
  • [Error]: Generating tiles failed

My questions are:

  1. How can I verify that my imported roll pitch heiding, orientations are correct in PIX4Dmapper?
  2. Does this failure mainly indicate insufficient longitudinal overlap between images?
  3. For a vehicle-mounted downward-looking road camera, are there recommended settings for dense matching or calibration?
  4. Is there a recommended workflow for projects without EXIF but with external position and orientation data?